We will use prostate cancer as a model to analyze variations in health resource utilization, cost of care, and factors associated with the process of chronic illness as antecedents and responses to health problems. Objective: to analyze the complex interplay of prostate cancer treatment, comorbidities, socioeconomic, hospital and physician factors and their relationships with treatment modalities, health resource utilization and direct medical care cost. Specific Aims (1): To analyze the variation in short and long term health resource utilization and direct cost of care across African American and Caucasian elderly prostate cancer patients, controlling for stage at diagnosis, physician and hospital characteristics and comorbidity. (2): To determine if differential rates of surgery (or radiation therapy) contribute to survival disparity between African American and Caucasian elderly prostate cancer patients. (3): To determine if differences in physician characteristics contribute to survival disparity between African American and Caucasian elderly prostate cancer patients. (4): To determine if differences in hospital characteristics contribute to survival disparity between African American and Caucasian elderly prostate cancer patients. Methods: We propose a cohort control design using SEER-Medicare databases. Prostate cancer patients diagnosed in 1995 or 1996 will be identified and followed retrospectively one-year pre and five years post diagnosis. Health resource utilization, treatment modalities and cost of care will be analyzed across ethnicity. Analyses will adjust for comorbidity, socioeconomic status and stage. To compute incremental cost attributable to prostate cancer, a control group of patients without cancer will be selected (matched by age, ethnicity and zip-code) from the Medicare database. Cumulative cost per patient per year will be computed and adjusted to 2005 dollars. For the prostate cancer group, total cost of care will be differentiated, as cost attributable to prostate cancer vs other diagnoses. Multivariate log linear regression models will be used to identify factors associated with direct medical cost across ethnicity. Cox proportional hazard models will be used to study the association between type of treatment and survival across ethnicity. We will use Instrumental variable approach to address the issues related to the selection bias.The outcome of this study will generate preliminary results and a working model leading to a RO1. [unreadable] [unreadable]